Cowboy Space is building orbital data centers that it eventually wants to launch on its own rockets, and that pitch just brought in $275 million in new funding. The bet is simple: AI compute demand keeps climbing, while power, land, permitting, and launch slots are all getting tighter. Founder Baiju Bhatt started the company in 2024 after leaving his day-to-day role at Robinhood, and he’s now taking the startup in a much more aggressive direction than its original space-solar plan.
That’s a huge swing.
And honestly, it has to be. If Cowboy Space wants to make orbital compute more than a science project, it can’t wait around for other people’s rockets.
What is Cowboy Space building and how would it work?
Cowboy Space is designing an orbital data center satellite that uses solar power in space, but the real twist is the vehicle itself. Bhatt says the company wants to turn the rocket’s second stage into the satellite, so the upper stage that reaches orbit becomes the compute platform instead of dead hardware. In practice, that means launching a booster, leaving the upper stage in orbit, and using that stage as the power-and-compute node for GPU workloads in space. Aetherflux — the company’s original name — still describes the concept as an orbital data center satellite that taps solar energy for AI’s growing power needs.
The hardware target is unusually concrete for a company this early. Cowboy Space expects each satellite to weigh 20,000 to 25,000 kilograms and generate 1 megawatt of power. It would support just under 800 onboard GPUs. Bhatt says the launcher would need to be a bit more capable than Falcon 9, though still smaller than Starship, and the long-term plan includes a reusable booster.
That design choice matters because it cuts out a lot of complexity. A general-purpose orbital launcher has to serve many payload types. Cowboy Space is only trying to launch one thing — its own data-center satellites. That should simplify integration and deployment logic. It also eases mass constraints. Explorer 1, the first U.S. satellite, was itself built into a rocket’s final stage.
The near-term use case is narrower than the big AI story implies. The first practical workloads are likely edge-processing tasks for space sensors rather than full-blown cloud replacement. That’s the sane version of the pitch. Put compute closer to space-based data sources. Reduce what has to be sent back to Earth. Learn how orbital infrastructure behaves before trying to build a giant off-world GPU farm.
Why did Baiju Bhatt turn Aetherflux into Cowboy Space?
From Robinhood to rockets
Bhatt isn’t a random rich founder playing astronaut. He co-founded Robinhood with Vlad Tenev in 2013, served as co-CEO until November 2020, then stayed on as chief creative officer until March 2024. He holds a B.S. in physics and an M.S. in mathematics from Stanford, and Robinhood says he had started 2 finance companies in New York before launching the brokerage app. Cowboy Space’s own background materials add another piece: Bhatt grew up obsessed with space, partly because his father worked as a research scientist at NASA Langley.
That history matters because this isn’t his first moonshot inside a hard-regulated industry. Robinhood scaled into a public company with more than 20 million customers, which at least shows Bhatt can recruit talent and raise capital. He can also ship through a lot of noise. None of that proves he can build a rocket engine. But it does explain why investors are willing to hear him out.
Why the company changed course
The startup began life as Aetherflux, with a different idea. Bhatt launched it in 2024 to collect solar power in orbit and beam electricity back to Earth. The company talked about supplying hard-to-reach places such as remote military bases, islands, and disaster-hit areas. Then the orbital compute opportunity pulled the company sideways. If you’re already generating power in space, using some of that electricity in orbit starts to look more practical than shipping all of it back down.
That pivot created a second problem. Launch supply.
Bhatt says he tried to find a version of the business that relied on outside launch providers while Cowboy only built satellites. He came away convinced that there won’t be enough launch capacity in the next 3 to 4 years for a scaled orbital data center business — and that first-party rocket companies are likely to prioritize their own payloads. So the company’s next pivot was the wild one: build the rocket too.
Fundraising, team, and the competition problem
That decision is what this new financing is really paying for. Cowboy Space said it closed a Series B at a $2 billion post-money valuation, led by Index Ventures, with Breakthrough Energy Ventures, Construct Capital, IVP, and SAIC also participating. The company had already raised $80 million from investors including Index, Breakthrough Energy Ventures, Andreessen Horowitz, and New Enterprise Associates. Bhatt expects the first launch before the end of 2028.
He’s staffing for that ambition. Cowboy Space has brought in Warren Lamont, a former Blue Origin propulsion engineer, and Tyler Grinnell, a former SpaceX launch director. It also plans to build its own rocket engine — the hardest, priciest subsystem in the whole stack. Testing, manufacturing, and launch facilities are still getting sorted out.
Competition is already intense, even though the market barely exists. Starcloud raised a $170 million Series A in March 2026 at a $1.1 billion valuation and already put an Nvidia H100 GPU in orbit in November 2025. Google’s Project Suncatcher is studying a solar-powered constellation that would use TPU chips and laser links, with economics that depend on launch costs falling below $200 per kilogram by the mid-2030s. Aethero has also put Nvidia hardware into space. Over all of this hangs SpaceX, which could become both infrastructure supplier and direct rival. Blue Origin is still trying to make New Glenn a dependable commercial workhorse.
Cowboy Space’s pitch against all of them is vertical integration and focus. Starcloud is progressing with orbital compute. Google is playing the long game. SpaceX and Blue Origin build rockets for many missions. Bhatt is arguing that a company built around one payload class — data-center satellites — can move faster on cost and architecture if it controls the launch system too.
Why does Cowboy Space’s $275M round matter?
This round matters because it changes the startup from an audacious satellite company into an audacious launch company.
That’s not a cosmetic shift. It rewrites the risk profile.
Before this, Cowboy Space could tell investors a familiar story: orbital power, edge compute, maybe a novel satellite platform. Now it’s promising launch independence and custom engines. It also wants reusable hardware and a vertically integrated route to space-based AI infrastructure. That’s far more capital intensive, but it also goes straight at the bottleneck Bhatt thinks will choke everyone else.
There’s also a strategic signal in who backed it. Index Ventures was already in. Breakthrough Energy Ventures stayed involved. The investor thesis isn’t just “AI needs more GPUs.” It’s that energy scarcity and launch scarcity are converging, and the company that solves both in one architecture could own a weird but valuable corner of future compute.
For customers, if this works, Cowboy Space could offer something terrestrial providers can’t: compute deployed where solar power is continuous and cooling conditions are different. Space-based sensors also wouldn’t need to dump raw data back to Earth first. If it doesn’t work, the failure will probably happen at the boring layer — propulsion, facilities, certification, launch cadence. That’s how most rocket stories end.
How big is the market behind Cowboy Space?
The macro tailwinds are real, even if Cowboy Space still has to prove the hardware. The World Economic Forum and McKinsey project the global space economy will grow from $630 billion in 2023 to $1.8 trillion by 2035. That doesn’t mean orbital data centers become mainstream. It does mean investors are looking at a much bigger commercial space stack than they were a few years ago.
And on the ground, the data-center crunch is getting uglier. JLL says AI training workloads require 10x the power density of traditional data-center uses. It also found that 57% of projects were delayed by 3 months or more in 2025, while average equipment lead times remain far above pre-2020 levels. In other words, the earthly alternative isn’t exactly frictionless.
That’s why this idea keeps resurfacing. Not because space is easy. Because the grid, permitting, and supply-chain math on Earth is getting harder fast.
Cowboy Space still sounds a little nuts. Bhatt would probably agree with that. But “space data center” has moved from sci-fi punchline to actual venture category in less than 2 years, and Cowboy Space just bought itself a shot at becoming the company that controls both the payload and the ride. The next thing to watch isn’t another slogan or render. It’s whether the startup can turn this financing into engine tests, facilities, and a credible path to a 2028 launch window.
Read how Helsing neared a $1.2B funding round to build AI-powered defense software and autonomous drones for Europe’s modern warfare systems.
FAQ
– What is the latest Cowboy Space funding round?
Cowboy Space has raised a new Series B that values the company at $2 billion post-money. Index Ventures led the round, and the capital is meant to help fund a much bigger push into rocket development, not just satellite work.
– How does Cowboy Space’s product actually work?
The company wants to launch orbital data center satellites that use solar power in space to run onboard compute. Its unusual idea is to build those compute platforms directly into the rocket’s second stage, so the upper stage becomes the satellite once it reaches orbit.
– Who founded Cowboy Space?
Baiju Bhatt founded the company after stepping away from his operating role at Robinhood in 2024. He co-founded Robinhood in 2013, studied physics and mathematics at Stanford, and has now turned his attention from fintech to launch systems and space-based power.
– Is Cowboy Space a rocket company or a space data center company?
It’s both now, and that’s the whole point. The company started as Aetherflux with a space-solar concept, then shifted toward orbital compute, and now says it needs its own launcher because commercial launch capacity is too scarce to scale a space data center business on someone else’s schedule.
